Genomic and Transcriptional Immunoediting of Acute Myeloid Leukemia in Response to Allogeneic Hematopoietic Stem Cell Transplantation

The curative potential of Allogeneic Hematopoietic Stem Cell Transplantation (HSCT) for Acute Myeloid Leukemia (AML) is mostly based on the ability of donor lymphocytes to actively control the outgrowth of host residual malignant cells. We demonstrated in the specific context of HLA-haploidentical HSCT that this equilibrium is frequently broken by de novo genomic mutations, causing loss of the mismatched HLA haplotype in leukemic cells, in turn leading to immune evasion and clinical disease relapse (Vago et al., N Engl J Med, 2009). This finding led us to hypothesize that different de novo genomic or transcriptional alterations may be at the base of other cases of post-transplantation AML relapse, prompting us to verify this hypothesis by high-throughput techniques in the present study.

Paired samples of AML harvested at diagnosis and at relapse after non T cell-depleted allogeneic HSCT (either from HLA-identical, matched unrelated or haploidentical donors) were FACS-purified and analyzed by whole-genome Single Nucleotide Polymorphysm (SNP) profiling (by the Illumina Human660W-Quad BeadChip) and gene expression analysis (by the Illumina HumanHT-12 v3.0 Expression Bead chip). The presence and allelic burden of the Internal Tandem Duplication (ITD) in the FLT3 gene was validated by a locus-specific PCR followed by capillary electrophoresis.

In 6 out of the 11 patients without genomic loss of HLA analyzed to date (54.5%), SNP profiling demonstrated the acquisition of de novo genomic alterations at post-transplantation relapse, preferentially occurring in well-characterized leukemia-associated genes (WT1, FLT3). Uniparental Disomy (UPD) was the most common alteration (45.5% of cases at diagnosis and 63.6% at relapse), frequently involving chromosome 13 in regions encompassing the FLT3 gene. Interestingly, in 5/11 patients (45.4%) we observed clonal evolution of leukemia at relapse from an oligoclonal population at diagnosis, with preferential expansion of the clones carrying genomic alterations. A total of 5 patients showed at relapse either de novo appearance of chromosome 13 UPD or the enrichment in a preexisting clone carrying the alteration: interestingly 4 out of these 5 patients gained in the process a substantial increase in the allelic burden of the ITD form of the FLT3 gene, which tightly correlates with aggressive behavior of leukemia.

Gene expression profiling of paired diagnosis-relapse samples from 7 patients revealed deregulation by at least three-fold in an average of 3% (0.8–5.1%) genes at post-transplantation relapse as compared to diagnosis. An unsupervised Gene Ontology analysis of these genes evidenced a significant enrichment in immune-related processes (p<10⁻¹⁴) in 3 of these patients, who displayed the earliest relapse onset after transplantation (median time 30 days, range 29–54). Such enrichment was not observed in the remaining 4 patients with later relapse onset (median time 152 days, range 98–299), suggesting a rapid imprint of the donor immune cells contained in the graft on the leukemia transcriptional profile.

A relevant observation was made in a patient who relapsed early after haploidentical HSCT without any documented genomic alteration in the HLA locus. In this patient, we could demonstrate at the time of post-transplantation relapse the specific transcriptional downregulation of the HLA class II antigen presentation pathway, whereas class I expression was preserved. Moreover, HLA class II expression was completely recovered in the absence of T cell immune pressure when the leukemic blasts harvested at relapse were transferred in immunodeficient NOD/SCID mice, suggesting plasticity of this mechanism of immune evasion.

Our data demonstrate that both genomic and transcriptional alterations occur frequently at leukemia relapse after allogeneic HSCT. The frequent selection of leukemic clones displaying specific unfavorable mutations (such as FLT3-ITD) from initially mixed populations grants a biological rationale for the post-transplantation use of targeted inhibitors to block the outgrowth of the more aggressive subsets. Moreover, the observation that not only genomic, but also transcriptional HLA loss can be at the basis of relapse after haploidentical HSCT warrants further optimization of current protocols of post-transplantation monitoring and treatment of disease relapse.

Bonini:MolMed SpA: Consultancy.